automated extraction of road surface information...
TRANSCRIPT
D r. J o n at han L i , P ro fe s sor
Fa c u l t y o f Env i ronment , U n i vers i t y o f Wate r loo, Ca na d a
S c h ool o f I n fo rm at ics , X i a m en U n i vers i t y, Ch i n a
ju n l i@uwater loo.ca, ju n l i@xm u.edu.cn
J u n e 1 7 , 2 0 1 4
AUTOMATED EXTRACTION OF
ROAD SURFACE INFORMATION
FROM MOBILE LIDAR
PRESENTATION OUTLINE
1. Introduction to Mobile LiDAR or MLS
2. Why Mobile LiDAR or MLS?
3. Road Surface Information Extraction
4. Concluding Remarks
5. Acknowledgements
6. Published Papers
CURRENT MLS SYSTEMS
• 3D Laser Mapping: StreetMapper (2005),
StreetMapper 360 (2011)
• Optech: Lynx Mobile Mapper (2007), Lynx
SG1 (2013)
• Riegl:VMX-250 (2009), VMX-450 (2011)
• SITECO: Road-Scanner (2009)
• Topcon: IP-S2 (2009), IP-S2 Compact+ (2012)
• Trimble: MX8 (2010)
• MDL Laser Systems: Dynascan (2010)
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
DIRECT GEOREFERENCING
Laser
Scanner
Xm
Ym
Zm
IMU P
GPS
antenna
M-frame
��� ��
� ������� ��
�
���� � � ��/��
� � ��/���
� ��� � ��/��
��� ��� � ��/�
��
� ���
�� ��
���
���
���
���
���
����
� ��/���
� ��/��
� ���
� ��/�� ��
� ��/� ��
�� �� ��
� The position of object P
in laser scanner
coordinate system
��� The position of
Object P in the local
coordinate system
��� The position and
orientation of laser
scanner in the local
coordinate system���
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
System Road Scanner IP-S2 MX8 VMX-450 StreetMa
pper 360
Dynascan Lynx
Scanner Faro Photon 120 Sick
LMS 291
VQ-450 MDL V100
Max. range 120m (ρ90%) 80m (ρ10%) 800m (ρ80%) up to 500m 200m (ρ80%)
Range
precision
1mm@ 25m, ρ90% 10 mm @ 20 m 5mm @150m (1σ) 8mm,1σ
Range
accuracy
±2mm@25m ±35mm 8mm @150m (1σ) ±5cm ±10mm (1σ)
PRR 122- 976 kHz 40kHz 2 x 550 kHz 36 kHz 2 x 500 kHz
Scan speed 48Hz 75Hz 2x 400 Hz 30 Hz 2x 100 Hz
Scanner FOV H360º / V320º 180º / 90º 360º without gaps 360º 360º
Angular
resolution
H0,00076°/
V0,009º
1º / 0,5º 0,001º 0,01º 0,001º
Weight 14.5 kg 22.7kg 11kg 11kg 78 kg
MOBILE LIDAR
MLS POINT CLOUD
Large Scale Unorganized 3D Point Cloud
Data Size
5GB per km
Density
2K pints / m2
Acquisition
30~100 km/h
Unorganized
Distribution
ACCURACY REQUIREMENTS
Reference: NCHRP 15-44 GUIDELINES FOR THE
USE OF MOBILE LIDAR IN TRANSPORTATION
APPLICATIONS.
The orders of accuracy:
1 = High (< 0.05 m)
2 = Medium (0.05 - 0.20 m )
3 = Low (> 0.20 m )
The levels of point density :
A = Fine (>100 pts/m²)
B = Intermediate (30 - 100)
C = Coarse (<30)
• Point cloud density (resolution) is determined by two factors:
• Measurement distance: 7000–8000 pts/m² (1 m), 800–900 (10 m), 80–90 (100 m), 50-60 (120 m), by a VMX-250 or MX8 at speed of 50 km/hr; 5000-6000 (1m), 400-500 (10m), 40-50 (100m), 20-30 (120m) at 120 km/hr.
• Driving speed: 0.15 m in scan line spacing at 50 km/hr, 0.35m at 120 km/hr.
RESOLUTION REQUIREMENTS
PROBLEMS FOR RAPID ACQUISITION OF
ROAD SURFACE INFORMATION
WHY MOBILE LIDAR OR MLS?
Sp
ee
d o
f da
ta
cap
ture
10 sq km/hr
0.1 sq km/hr
Capital cost
€400-500k
€100k €600k
Terrestrial
Laser
Scanning
(TLS)
Airborne
Laser
Scanning
(ALS)Mobile
Laser
Scanning
(MLS)
TRANSPORTATION APPLICATIONS
OF MOBILE LIDAR• Roadways
Road topo for design
Intersections
Pavement QA
Road topo for problem analysis
Paving volumes
Input to road milling
Accident investigation & analysis
Slope stability & retaining wall surveys
Toll Plazas
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
TRANSPORTATION APPLICATIONS
OF MOBILE LIDAR• Bridges and elevated roads
Design as-builts
Clearances
Topo for problem analysis
Heritage
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
TRANSPORTATION APPLICATIONS
OF MOBILE LIDAR• Tunnels
Profiles
Pavement QA & quantities
Clearances
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
RESULTS FOR ROAD SURFACE
INFORMATION EXTRACTION
• Road information (road markings, pavement
crakes, manholes, etc.)
• Non-road information (light-poles, trees, cars,
power-lines, etc.)
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD SURFACE EXTRACTION
1) Point cloud data profiling
2) Profile gridding and principal point generation
3) Curb corner point detection
4) Road edge interpolation and road surface extraction
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
PROFILING & CURB CORNER
POINT DETECTION
(a) Data profiling model;
(b) Profile generation in real point clouds;
(c) Generated profiles;
Profile gridding and principal point generation
Detected curb corner points
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
CURB-LINE INTERPOLATION &
ROAD SURFACE EXTRACTION
Curb corner points from all profiles.
Interpolated curb-lines.
Extracted road surfaces
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD SURFACE EXTRACTION
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
2D ROAD MARKING EXTRACTION
1. Generation of
geo-referenced
intensity image
2. Thresholding
3. Extraction of
road markings
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
GEO-REFERENCED INTENSITY
IMAGE GENERATION
Geo-referenced intensityimage generation model
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING EXTRACTION
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING EXTRACTION IN 3DRoad surface extraction
Extracted road surface points
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING POINT EXTRACTION
Road marking points extraction using multi-segment thresholding
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING CLASSIFICATION
Road marking points clustering
Detected large-size road markings
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING CLASSIFICATION
Large-size road marking classification
Normalized cut segmentation on connected road markings
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING CLASSIFICATION
Deep learning based small-size road markingclassification
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING CLASSIFICATION
PCA-basedzebra crossing anddashed lineclassification
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MARKING CLASSIFICATION
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
PAVEMENT CRACK EXTRACTION
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
CRACK EXTRACTION RESULTS
(a) Geo-referenced intensity image, (b) extracted cracks, and (c) overlaid result.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MANHOLE DETECTION
(a) Extracted road surface and (b) geo-referenced intensity image.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MANHOLE DETECTION
Road manhole detection based on marked point process of rectangles and disks.
Mark point of disk. Mark point of rectangle.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MANHOLE DETECTION
Transformations of mark points.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ROAD MANHOLE DETECTION
(a) Geo-referenced intensity image and (b) road manhole detection result.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
CONCLUDING REMARKS
• With a MLS system, mobile mapping crew can drive a
highway, rural road, railroad, or on the shoreline of a river
or lake.
• Along the way, the system captures virtually anything
visible to the eyes in 3D. The collected data are a totally
immersive 3D view of objects and surroundings.
• Today's major trend in mapping and GIS is an increasing
demand for not only accuracy of geospatial data but
efficiency and low cost.
• MLS systems can meet this demand and provide the end
results with increased productivity.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
CONCLUDING REMARKS• MLS is a much safer mapping technique than traditional
highway surveys, where surveyors wearing orange vests
measure the land boundaries and understand the terrain via
total stations, TLS, and so on, as well as the requirement of
extensive traffic management or road closures.
• MLS is a more feasible 3D measurement technology for
large-scale mapping projects than the legacy methods.
• Specifically speaking, a 10-km-long highway would have
taken at least 20 nights to survey and a week to process the
resultant measurements by a traditional highway survey
method, while the highway, would take, from start to finish,
less than a week using a MLS system.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
ACKNOWLEDGEMENTS
• NSERC and NSFC for funding support.
• UW and XMU for support;
• GeoSTARS Lab at UW and Fujian Key
Laboratory of Sensing and
Computing for Smart Cities (SCSC
Lab) at XMU for support;
• Special thanks go to Dr. Haiyan Guan
and Yongtao Yu.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
PUBLISHED JOURNAL PAPERS① Guan, H., J. Li, Y. Yu, and C. Wang, 2014. Automatic road information extraction using
mobile laser scanning data. IEEE Transactions on Intelligence Transportation Systems,
Doi:10.1109/TITS.2014.2328589.
② Guan, H., J. Li, Y. Yu, and Wang, C., 2014. Interactive tensor voting method for crack
detection using mobile laser scanning data. IEEE Transactions on Geoscience & Remote
Sensing, accepted.
③ Guan, H., J. Li, Y. Yu, Wang, C., Chapman, M., and Yang, B., 2014. Using mobile laser
scanning data for automated extraction of road markings. ISPRS Journal of
Photogrammetry & Remote Sensing, 87:93-107.
④ Yu, Y., J. Li, H. Guan, C. Wang, 2014. Automated detection of road manhole and sewer well
covers from mobile LiDAR point clouds, IEEE Geoscience and Remote Sensing Letters, 11(9):
1549-1553.
⑤ Yu, Y., J. Li, J. Yu, H. Guan, C. Wang, 2014. Pairwise three-dimensional shape context for
partial object matching and retrieval on mobile laser scanning data, IEEE Geoscience and
Remote Sensing Letters, 11(5): 1019-1023.
⑥ Yu, Y., J. Li, H. Guan, C. Wang, 2013. A marked point process for automated building
detection from lidar point-clouds, Remote Sensing Letters, 4(11): 1127-1136.
⑦ Wang, H., C. Wang, H. Luo, P. Li, M. Cheng, C. Wen, J. Li, 2014.
Object detection in terrestrial laser scanning point clouds based on Hough Forest, IEEE
Geoscience and Remote Sensing Letters, 11(10): 1807-1811.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
PUBLISHED CONFERENCE PAPERS
① Guan, H., J. Li, Y. Zhou, Y. Yu, C. Wang, M.A. Chapman, 2014. Automatic extraction
of power lines from mobile laser scanning data, IGARSS 2014, Quebec City, Quebec,
July, 4p.
② Jia, F., J. Li, C. Wang, Y. Yu, M. Cheng, D. Zai, 2014. Earthwork volumes estimation in
asphalt pavement reconstruction using a mobile laser scanning system, IGARSS 2014,
Quebec City, Quebec, July, 4p.
③ Yu, Y., J. Li, H. Guan, C. Wang, 2013. Automated detection of road manhole covers
from mobile LiDAR point-clouds based on a marked point process, GiT4NDM,
Mississauga, Ontario, October 9-11, 4p.
④ Guan, H., J. Li, and Y. Yu, 2013. Rapid update of road surface databases using
mobile LiDAR, GiT4NDM, Mississauga, Ontario, October 9-11, 4p
⑤ Li, J., Y. Yu, H. Guan, C. Wang, 2013. Extraction of tree crowns from mobile laser
scanning data using a marked point process model, MMT2013, Tainan, Taiwan, 6p.
⑥ Yu, Y., J. Li, H. Guan, C. Wang, 2013. Detection of road surface cracks from mobile
laser scanning data, MMT2013, Tainan, Taiwan, 6p. (Best Student Paper Award)
⑦ Guan, H., J. Li, Y. Yu, C. Wang, 2013. Geometric validation of a mobile laser scanning
system for urban applications, MMT2013, Tainan, Taiwan, 6p.
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014
XXV International Federation of Surveyors Congress, Kuala Lumpur, Malaysia, 16 – 21 June 2014